metadata
license: mit
base_model: HuggingFaceH4/mistral-7b-sft-beta
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-full
results: []
zephyr-7b-dpo-full
This model is a fine-tuned version of HuggingFaceH4/mistral-7b-sft-beta on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0984
- Rewards/chosen: -1.3191
- Rewards/rejected: -2.1712
- Rewards/accuracies: 0.7695
- Rewards/margins: 0.8521
- Logps/rejected: -474.4743
- Logps/chosen: -388.9529
- Logits/rejected: -2.3033
- Logits/chosen: -2.3263
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 2
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.1368 | 0.21 | 100 | 0.1234 | -0.7206 | -1.1646 | 0.6953 | 0.4441 | -373.8169 | -329.0978 | -2.7113 | -2.7294 |
| 0.0936 | 0.42 | 200 | 0.1059 | -1.0413 | -1.7570 | 0.7422 | 0.7157 | -433.0510 | -361.1696 | -2.4844 | -2.4997 |
| 0.1045 | 0.63 | 300 | 0.1050 | -1.1721 | -1.9852 | 0.7734 | 0.8130 | -455.8698 | -374.2533 | -2.3263 | -2.3482 |
| 0.1007 | 0.84 | 400 | 0.0984 | -1.3191 | -2.1712 | 0.7695 | 0.8521 | -474.4743 | -388.9529 | -2.3033 | -2.3263 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1